Adsorption and surface reactions of hydrogen sulfide and sulfur

K. T. Leung, X. S. Zhang, and D. A. Shirley. J. Phys. Chem. , 1989, 93 (16), pp 6164–6170. DOI: 10.1021/j100353a042. Publication Date: August 1989...
0 downloads 0 Views 924KB Size
6164

J . Phys. Chem. 1989, 93, 6164-6170

adsorption sites, while adsorbed 0 has no effect on the methanSome details of the reaction mechanism have been elucidated. ethiol sticking coefficient. Several researchers have found that Strong evidence for the formation of methyl thiolate on this surface S adsorbs in the 2-fold hollow sites on the Ni( 1 IO) ~ u r f a c e , ~ ~ - ~ *has been obtained from high-resolution electron energy loss and X-ray photoelectron spectroscopies and from deuterium incorand the competition for sites suggests that the thiolate may adsorb poration experiments. The methyl thiolate species is formed upon in the same site. Oxygen, however, does not appear to compete for adsorption sites which is consistent with its location in bridge adsorption and is stable until about 210 K when the shift in the sites on a reconstructed Ni ~ u r f a c e . ~ ~ - ~ ’ S 2p binding energy indicates that C-S bond scission occurs. Methane desorption commences at about the same temperature, Conclusions prohibiting the differentiation between a mechanism involving Desulfurization of methanethiol to form methane and a surface direct hydrogenolysis of a thiolate and a mechanism entailing sulfide has been achieved on Ni(ll0) surfaces at high coverages. formation and hydrogenation of surface methyl species. At low coverages, the reaction of methanethiol on the Ni surface The effects of electronegative surface modifiers, sulfur and results in decomposition to form surface sulfur, carbon, and hyoxygen, have been studied. The major effect was to enhance the drogen; the latter desorbs as H2. rate of methane formation relative to total decomposition. This results from either electronic or steric factors. (33) Baudoing, R.; Gauthier, Y.; Joly, Y. J. Phys. C 1985, 18, 4061. (34) Warburton, D. R.; Thornton, G.; Norman, D.; Richardson, C. H.; McGrath, R.; Sette, F. Surf.Sci. 1987, 1891190, 495. (35) Ohta, T.; Kitajima, Y.; Stefan, P. M.; Shek Stefan, M. L.; Kosugi, N.; Kuroda, H. J. Phys. 1986, 47, 503. (36) Fauster, T.; Durr, H.; Hartwig, D. Surf.Sci. 1986, 178, 657. (37) Van der Veen, J. F.; Tromp, R. M.; Smeenk, R. G.; Saris, F. W. Surf. Sci. 1979, 82, 468. (38) Robey, S. W.; Barton, J. J.; Bahr, C. C.; Liu, G.; Shirley, D. A. Phys. Reu. B 1987, 35(3), 1108. (39) Bard, A. M.; Ollt, L. Surf.Sci. 1983, 126, 170. (40) Niehus, H.; Coma, G. Surf.Sci. 1985, 151, L171. (41) Bar& A. M.; Binnig, G.; Rohrer, H.; Gerber, C.; Stoll, E.; Baratoff, A.; Salvan, F. Phys. Reu. Lett. 1984, 53, 1304.

Adsorption and Surface Reactions of Energy Loss Spectroscopy

).I$

Acknowledgment. Research was sponsored by the Division of Chemical Sciences (Office of Basic Energy Sciences), U S . Department of Energy, under Contract No. DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc. D.R.H. thanks D. M. Zehner and G. W. Ownby for the Ni single crystal and for the use of the X-ray source and S. H. Overbury, D. R. Mullins, and E. H. Taylor for many useful comments and suggestions through the course of this work. Registry No. Ni, 7440-02-0; 02, 7782-44-7; S, 7704-34-9; CHSSH, 74-93-1; H20, 7732-18-5; H, 1333-74-0; CH$, 82-8.

17302-63-5; CHI, 74-

and SO2 on Cu(100) Studied by Electron

K. T. Leung,+ X. S. Zhang,f and D. A. Shirley* Department of Chemistry, University of California, and Materials and Chemical Sciences Division, Lawrence Berkeley Laboratory, 1 Cyclotron Road, Berkeley, California 94720 (Received: January 3, 1989)

The adsorption of HIS and SOz on Cu(100) was investigated by use of electron energy loss spectroscopy as a function of coverage, temperature, and scattering angles. In particular, irreversible dissociation of H2S and the formation of sulfhydryl (SH) species at low and immediate coverage on Cu(100) were observed at low temperature. This was followed by molecular physiadsorption at higher coverage. In the case of SOz on Cu( loo), decomposition of SO2and the formation of SO3 surface species at room temperature were observed. A surface reaction involving preadsorbed SOzon Cu(100) with H2S is discussed.

I. Introduction Since the pioneering work of Propst and Piper,’ electron energy loss spectroscopy (EELS)Zhas become a popular, nondestructive technique for surface analysis when high surface sensitivity is desired. Detailed information concerning the bonding of various surface adsorbates and their intermediates can be obtained from their characteristic vibrational Many surface intermediates,*” as well as their adsorption and desorption behavior, have been identified by EELS. Furthermore, recent developments in electron optics have allowed the study of the kinetics of surface reactions involving these surface species.12 The adsorption behaviors of H2S and SOz on surfaces have attracted a great deal of attention because of the industrial and environmental importance of these molecules. Adsorption properties of H2S on Rh( lOO),I3 Mo( 100),14 MoSZ,l5Si( 11 1),I6 and GaAs” have been investigated by a variety of techiques including low energy electron diffraction (LEED), photoemission, thermal Present address: Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada. *Permanent address: Department of Physics, Zhejiang University, Hongzhou, People’s Republic of China.

0022-3654/89/2093-6164$01.50/0

desorption, and EELS. Of particular interest are the photoemission studies of H2S on Ru(1 lo),’* W(lOO),I9 and GaAs sur(1) Propst, F. M.; Piper, T. C. J. Vac. Sci. Technol. 1967, 4 , 53.

(2) Ibach, H.; Mills, D. L. Electron Energy Loss Spectroscopy and Surface Vibrations; Academic Press: New York, 1982. (3) Andersson, S.; Harris, J. Phys. Reu. Lett. 1982, 48, 545. (4) Persson, M.; Andersson, S.; Karlsson, P. A. Chem. Phys. Lett. 1984, 111, 597. (5) Sexton, B. A. J. Vac. Sci. Techonol. 1979, 16, 1033. (6) Andersson, S.; Nyberg, C.; Tengstal, C. G. Chem. Phys. Lett. 1984, 104, 305. (7) Andersson, S.; Persson, B. N. J.; Persson, M.; Lang, N. D. Phys. Reu. Lett. 1984, 52, 2073. (8) Persson, M.; Andersson, S . SurJ Sci. 1982, 117, 352. (9) Sexton, B. A. Surf.Sci. 1979, 88, 319. (IO) Persson, B. N. J.; Ryberg, R. Phys. Reu. Lett. 1982, 48, 549. ( 1 1 ) Sexton, B. A. Surf.Sci. 1979, 88, 299. (1 2 ) Kevan, S. D.; Dubois, L. H. Reu. Sci. Instrum. 1984, 55, 1604. Ho, W. J. Vac. Sci. Technol. 1985, A3, 1432, and references therein. (13) Hegde, R. I.; White, J. M. J. Phys. Chem. 1986, 90, 296. (14) Wilson, J. M. Surf. Sci. 1976, 57, 499. Clarke, L. J. Surf.Sci. 1981, 102, 331. (15) Salmeron, M.; Somorjai, G. A,; Wold, A,; Chianelli, R.; Liang, K. S. Chem. Phys. Lett. 1982, 90, 105. (16) Fujiwara, K.; Ogata, H. Surf. Sci. 1978, 7 2 , 157.

0 1989 American Chemical Society

EELS of H2S and SO2 on Cu(100)

The Journal of Physical Chemistry, Vol. 93, No. 16, 1989 6165

faces,” and the electron energy loss studies of H2Son Ni(100)20 and Pt( 11 1),21which reported the formation of sulfhydryl (SH) species, at low temperature in most cases.18,20*21 Two different interpretations of the corresponding vibrational features were also presented in the latter two EELS studies.20-21 Similarly, a number of studies involving different techniques were reported for the adsorption of SO2and its reactions on metal surfaces. Blyholder and Cagle showed that the reaction of SO2 on Fe and Ni produced SO4surface speciesz2 A similar reaction involving SO4 as a product on Ag was also reported by Barber et al. using secondary ion mass s p e c t r o ~ c o p yand ~ ~ by Lassiter using Auger electron spectroscopy and work function measurem e n t ~ .Low, ~ ~ Goodsel, and Takezawa, on the other hand, studied the adsorption of SO2 on CaOZSand on Mg026 at room temperature using Fourier transform infrared spectroscopy and 0 1000 2000 3000 suggested SO3 as a possible surface species upon adsorption. Energy Loss (cm-’) Kohler and WassmuthZ8investigated the kinetics of SO2 on PtFigure 1. EEL spectra of a saturation coverage of HIS on Cu(100) (1 11) using a molecular beam technique and thermal desorption expose and recorded at (a) 100 and (b) 300 K. The fwhm’s of the elastic measurements and further suggested that SO2 adsorbed dissopeaks and the peak locations are indicated in cm-I. ciatively at low temperature for small coverages. Dissociative adsorption involving a wide range of possible surface species was equipped with retarding-field optics capable of LEED and Auger also observed on other surfaces, including Rh( 1 Pt( 1 electron spectroscopy (AES) as well as an ion sputtering gun for near liquid nitrogen temperatures. Pt( 11 1),28 and Pd( sample cleaning. A detailed description of our modified electron However, a recent study by Hussla and Heidberg,27using farenergy loss spectrometer was given elsewhere.33 Briefly, it and mid-infrared transmission spectroscopy, found no evidence consisted of a 2-in. mean diameter double-pass hemispherical of dissociative adsorption for SO2 on Cu(100) over a wide temperature range. On Ag( loo), Ag( 11 l ) , Ag( 1 and A U , ~ ~ monochromator and a 5-in. mean diameter single-pass hemispherical analyzer. The analyzer was rotatable in both polar and molecular adsorption at low temperature and dissociative adazimuthal directions with respect to the normal of the crystal to sorption at higher temperature31 were reported. facilitate off-specular measurements without rotating the sample. An earlier publicationz0 from this laboratory gave a detailed Two separate three-aperture lenses were used as the output optics characterization of the adsorption and decomposition of H2Son for the monochromator and as the input optics for the analyzer. Ni(100). In this report, we present a parallel study on the adIn the earlier design, the double hemispheres of the monochrosorption of H2S and SO2on Cu( 100) using EELS. Surface species mator were separated by a single aperture. In the present work, generated from the adsorption and desorption of HIS and SO2 the aperture was replaced by a three-aperture lens to enable the on Cu( loo), as well as from reactions between H2Sand a Cu(100) two hemispheres to operate at different pass energies.34 This surface preadsorbed with SO2, were characterized using the modification improved the signal rate by a factor of 5-10 while measured electron energy loss spectra. Except for the differences maintaining similar energy (40 cm-’ fwhm) and angular (3O in the values of the vibrational frequencies, very similar adsorption half-angle) resolution. and decomposition behaviors can be inferred from the vibrational A high-purity Cu crystal was cut, polished, and oriented to spectra associated with these surface species on Cu( 100) and 1 O from the (100) plane by standard methods. It was (pre-) within Ni( 100)20in the case of H2S. For SO2,similar adsorption patterns cleaned by a well-known procedure: repeated cycles of Ar ion were found on Cu( 100) and Ag( 1 sputtering at 1 keV followed by annealing at 800 K for 15 min. The cleanliness of the sample was checked by use of retarding-field 11. Experimental Section AES. The sample could be cooled to liquid nitrogen temperature The angle-resolved electron energy loss spectrometer was housed or heated to above 1500 K by use of electron beam bombardment in an UHV chamber (base pressure = 1 X Torr), which was from the back of the sample plate. The temperature of the crystal was measured by using a Chromel-Alumel thermocouple spotwelded on the sample plate 7 mm above the crystal. The accuracy (17) Massies, J.; Etienne, P.; Dezaly, F.; Linh, N. T. Surf. Sci. 1980, 99, of the temperature measurements was estimated to be f20 K. 121. Ranke, W.; Kuhn, H. J.; Finster, J. Surf. Sci. 1987, 192, 81, and Detailed descriptions of the effusive beam dosing procedure as references therein. (18) Fisher, G. B. Surf. Sci. 1979, 87, 215. well as the calibration procedure using an ambient dosing tech(19) Bhattacharya, A. K.; Clarke, L. J.; De La Garza, L. M.; J . Chem. nique were given e l s e ~ h e r e . ~Exposures ~ , ~ ~ of H2S (Matheson, Soc., Faraday Trans. I 1981, 77, 2223. 99.5% purity) or SO2 (Matheson, 99.98% purity) was achieved (20) Baca, A. G.; Schulz, M. A.; Shirley, D. A. J. Chem. Phys. 1984, 81, by maintaining an ambient chamber pressure of 1 X Torr 6304. (21) Koestner, P. J.; Salmeron, M.; Kollin, E. B.; Gland, J. L. Chem. Phys. with the adsorbant for a specified period of time with the sample Lett. 1986, 125, 134. positioned at 1 cm from an effusive beam doser. Exposures (22) Blyholder, G. D.; Cagle, G . W. Enuiron. Sci. Technol. 1971, 5, 158. obtained by this beam dosing technique were calibrated with those (23) Barber, M.; Sharp, P.; Vickerman, J. C. Chem. Phys. Lett. 1974.27, obtained by using an ambient dosing procedure, by comparing 436. (24) Lassiter, W. S . J . Phys. Chem. 1972, 76, 1289. the corresponding S(152 eV) t o Cu(62 eV) relative Auger in(25) Low, M. J. D.; Goodsel, A. J.; Takezawa, N. Enuiron. Sci. Technol. Toms exposure using the beam dosing tensities. A 1 X 1971. ~ 5. , . 1191. , procedure corresponded to 0.5-1 langmuir (1 langmuir = 1 X lo4 (26) Goodsel, A. J.; Low, M. J. D.; Takezawa, N. Enuiron. Sci. Technol. Torr-s). A saturation coverage was obtained after 40- and 201972, 6, 268. (27) Hussla, I.; Heidberg, J. J . Electron Spectrosc. Relat. Phenom. 1986, langmuir exposures of H2S and SO2, respectively. A new sample 39, 213. was prepared for each exposure, to avoid accumulative contam( 2 8 ) Kohler, U.; Wassmuth, H. W. Sur/. Sci. 1983, 126, 448, and referination. ences therein. All specular EELS measurements were made at an incident (29) Ku, R. C.; Wynblatt, P. Appl. Surf. Sci. 1981, 8, 250. (30) Burke, M. L.; Madix, R. J. Surf. Sci. 1988. 194. 223. angle of 55O from the surface normal, while off-specular mea~

~~

~

(31) Outka, D. A.; Madix, R. J.; Fiiher, G. B.; Maggio, C. D. J . Phys. Chem. 1986, 90,4051. Rouida, R.; Pratesi, F. Surf. Sci. 1981, 104, 609, and references therein. (32) Brundle, C. R.; Carley A. F.; Faraday Discuss. Chem. Soc. 1975,60, 51. Furuyama, M.; Kishi, K.; Ikeda, S . J . Electron Spectrosc. Relat. Phenom. 1978, 13, 59.

(33) Schulz, M. A. MSc. Thesis, University of California, Berkeley, 1985, LBL-19769. A. G. Baca, Ph.D. Thesis, University of California, Berkeley, 1985, LBL-19099. (34) Kesmodel, L. L. J . Vac. Sci. Technol. 1983, A I , 1456.

6166

The Journal of Physical Chemistry, Vol. 93, No. 16, 1989

Leung et al.

TABLE I: Comparison of Vibrational Frequencies (f4cm-') of

Cu( 100)

Adsorbed Hfi and D2S on Cu(100) at 100 K with Gas-Phase Data3s

54\\'

---

~

( a ) H,S 1144

i,i 0

I

,\ 1000 2000 Energy Loss (cm-')

adsorbed on Cu(100) at 100 K

gas-phase

( b ) D,S

2469

x 1000 3000

vibration H-S stretch asymmetric symmetric H-S bend H-Cu stretch H-S stretch or bend S-Cu stretch phonon or hindered translations

(a) H2S

D2S

(b)

ratio (a/b)

2684 2611 1290

1999 1892 934

1.34 1.38 1.38

(c) HIS 2469

(d) ratio D2S (c/d)

1144 550 430 320 178

859 452 317 317

Figure 2. EEL spectra of a saturation coverage of (a) H2S and (b) D2S on Cu(100) at 100 K .

H,S/Cu( 1 OOK

surements were conducted by rotating the analyzer through a fixed angle from the specular direction, away from the surface normal. An impact energy of 1.6 eV was used for all measurements. Individual peaks in the EEL spectra were deconvoluted by using a combination of Gaussians as well as a background function to approximate the tail of the elastic peak. The accuracy of the reported loss frequencies so determined was estimated to be & 4 cm-l.

(c)

10' off-specular

(b)

5" off-specular

(a)

specular

111. Results A . EELS of H2S/Cu(100). Figure 1 shows the EEL spectra of a Cu( 100) surface exposed to a saturation coverage of H,S and recorded at (a) 100 K and (b) 300 K in the specular direction. A diffuse ~ ( 2 x 2 LEED ) pattern with streaks between the quarterand half-order spots was observed at both temperatures, with the pattern obtained at 100 K being more diffuse. Further annealing of the sample at 300 to ca. 500 K for 10 min produced a sharp ~ ( 2 x 2 LEED ) pattern with no observable streaks. Good agreement is observed between the reported p(2X2)S/Cu( 100) spectrum* and the room temperature spectrum (Figure lb). In particular, the dissociation of H2Son Cu(100) is indicated by the strong S-Cu stretch observed at 324 cm-'. The small downward shift (ca. 12 cm-I) relative to the reported S-Cu stretch* can be attributed to possible disorder effects for the dissociated species on Cu( 100) at room temperature. There is also some evidence of phonon vibrations,' as indicated by the weak shoulder at 177 cm-I. The spectrum at 100 K (Figure la) shows four additional peaks a t 430, 550, 1144, and 2469 cm-'. The two higher vibrational features at 1144 and 2469 cm-I can be readily identified respectively as H-S bending and stretching modes of molecularly adsorbed H2S by a comparison with the available vibrational data of free HIS molecules.35 In the gas-phase data the 1290- and 26 1 1 (2684-) cm-' peaks were assigned to H-S bending and symmetric (asymmetric) stretching modes, respectively. The feature at 178 cm-' in the 100 K spectrum can be attributed to either surface phonon vibration as in the room temperature case or hindered translations of molecularly adsorbed H2S. In order t o identify the remaining features (Le., at 430 and 550 cm-I), an EEL spectrum of a saturation coverage of D,S on Cu(100) at 100 K was measured (Figure 2b). The effect of isotopic substitution is to scale the frequencies of the H-related vibrations of H,S/Cu(100) with respect to those of the corresponding vibrations of D2S/Cu(100) by a factor of 21/2. The frequencies of these two systems as well as the gas-phase data are given in Table I. The observed D2S/Cu(100) features are consistent with the isotopic substitutional effect, with the scaling factors lying between 1.2 and 1.4. The D-S stretching modes of the molecularly adsorbed D2S were not observed, presumably due to their weak intensities. The broad structure at 317 cm-' may

-

( 3 5 ) Schimanouchi, T. Tables of Molecular Vibrational Frequencies; National Bureau of Standards, US.Department of Commerce: Washington, DC, 1972; Vol. 1.

0

1000 2000 Energy Loss (cm-')

1.33 1.22 1.36

100)

3000

Figure 3. EEL spectra of a saturation coverage of HIS on Cu(100) at 100 K as a function of scattering geometries: (a) specular at 5 S 0 , (b) 5' off-specular, and (c) IOo off-specular (note the change in the scaling factor for (c) only). The scaling factors of individual parts of the spectra are with respect to the corresponding elastic intensities, which are not normalized to each other.

be due to an overlap of two features: the scaled analogue of the H2S feature at 430 cm-I and the S-Cu stretch. The features at 430 (317) and 550 (452) cm-' for H&S(D2S) are therefore related to surface vibrations involving H (D) atoms. Figure 3 shows the EEL spectra of a saturation coverage of H2S on Cu(100) at 100 K measured in the specular and two off-specular scattering geometries. Assuming that the S-Cu stretch (at 320 cm-I) is dipole active, one can obtain a qualitative idea of the dipole (infrared) activity of other structures by comparing the changes in their intensities relative to the S-Cu stretch as a function of angle. This comparison confirms that the H-S bend (at 1144 cm-I) and H-S stretch (at 2469 cm-I) of molecularly adsorbed H2S have nondipole (Le., impact-scattering) contributions, as indicated by a large increase in the relative intensities in the 10' off-specular spectrum (note scale change in Figure 3c). In other words, H2S is adsorbed on the Cu surface such that these vibrations have parallel components in the dipole moments with respect to the surface. Furthermore, the feature at 430 cm-I, like the S-Cu stretch (at 320 c d ) , is dipole active. Together with the evidence obtained in Figure 2, we may identify this structure to be H-S stretch of a sulfhydryl (SH) species with S bonded in a head-on configuration perpendicular to the surface. Finally, the angular variation of the remaining feature at 550 cm-l is similar to that of the H-S stretch of the dissociated (SH) species. This may likely be due to a second H-S vibrational mode or a H-Cu vibration. Further evidence is required to distinguish between the two possibilities. A temperature-dependent study is shown in Figure 4. In this experiment, the Cu(100) surface was first exposed with a saturation coverage of H2S at 100 K (Figure 4a). I t was subsequently annealed to the specified temperatures for 5 min. The sample was then recooled back t o 100 K, at which temperature the EEL spectrum was recorded. I t should be noted that a new sample

EELS of H2S and SOz on Cu(100)

The Journal ofPhysica1 Chemistry, Vol. 93, No. 16, 1989 6167

TABLE 11: Comparison of Vibrational Frequencies ( 1 4 cm-I) of Free SO2, SO2 Adsorbed on Metal, and Other Oxides of SulfuP frequencies, cm-' systems a b C d e f 9; ref SO2/CU(100) this work at 100 K 290 346 515 978 1290 296 340 482 613 859 at 300 K 31

SOdAg(110) at 241 K at 418 K SO,/Ag(110) at 570 K at 815 K SO,/Pd( 100) at 115 K at 371 K

220 245

585 585

835 805

885 875

220 225

605 610

910 915

1070

1275 1285

1155 1155

1355 1250

1135 1143 1119 1151 1147

1304 1297 1259 1362 1308, 1330

1010 1069 1110

1330

1180 31 30

545 525

865

so2/cscI

27

a (220-240 K)

302 215 305

(3 (146-205 K) y (>200 K) SO2 gas SO2 solid sOJ2-

526 541 537 518 521

pyramidal planar

46 1 532 454

so42-

633 652 622

961 983

35 38 39 39

'Tentative assignments: b, metal-S stretch; c and d, S-0 bend; e-g, S - 0 stretch. 1

'

"

X150 I

1

~

'

~

1

"

H,S/Cu( 1 OOK

'

1

100)

Il1 4 4

kUI (a)

0

lOOK

1000 2000 Energy Loss (cm-')

3001

Figure 4. EEL spectra of a saturation coverage of H2S on Cu(100) at 100 K after different temperature treatments. The annealing tempera-

ture (in K) and time (in seconds) are indicated in each spectrum. was prepared for each run to avoid accumulative contamination. Figure 4b shows that the anneal at 130 K removed the two molecular features at high frequency as well as the structure at 430 cm-I. The removal of the two high-frequency features suggests that HzS was physisorbed on Cu(100) below 130 K. Furthermore, the dissociated SH species, which gave rise to the feature at 430 cm-I, was weakly adsorbed. The marked increase in the intensities of the S-Cu stretch and of the vibration at 550 cm-l (Figure 4b) further suggests that the SH species may (1) dissociate to form S(abs) and H(abs), and/or (2) rearrange itself to a different site or a different vibrational mode. Finally, further annealing to higher temperatures resulted in a gradual decrease in the intensities of the remaining features at 320 and 550 cm-I. A significant depletion in the intensity of the 178-cm-I peak after the 130 K anneal was also observed, suggesting that the 178-cm-' peak may have a contribution from the hindered translations of molecularly adsorbed H2S. Its residual intensity after the 130 K anneal (Figure 4c-e) can be attributed to phonon vibration. Qualitatively, we may conclude that the adsorption strength of the observed features increases in the following order: (178, 430, 1144, 2469 cm-l) (weak) < 550 cm-l < 320 cm-' (strong).

Energy Loss

(cm-')

Figure 5. EEL spectra of H2S on Cu(100) at 100 K as a function of

coverage. The number of seconds of exposure using an effusive beam dosing method is indicated in each spectrum. A 10-s exposure by this procedure corresponds to 0.5-1 langmuir (1 langmuir = 1 X lod Tows). The spectrum corresponding to a saturation coverage (-40 langmuirs) is given in (e). Finally, we present a coverage-dependent study (using a beam dosing p r o c e d ~ r e )to~ investigate ~-~~ the early stages of HzS adsorption on Cu(100) in Figure 5. The most striking change is the red shift of the feature at 600 cm-I at low coverage (Figure sa) to 550 cm-' at saturation coverage (Figure 5e). There is no other noticeable change in the frequencies of the other features with coverage. At low coverages (Figure 5a,b), H2S appears to dissociate into S and a species (most likely H) that gives rise to the feature at 550 cm-l at saturation coverage. These dissociated species saturate at an exposure of about 20 s (1-2 langmuir) before the SH species (at 430 C M - ~and ) the molecularly adsorbed HzS begin to accumulate. This is consistent with the order of adsorption strength of individual species asserted in the temperature-dependent study (Figure 4) above. Furthermore, it confirms that HzS dissociates to S (and H) at low coverage on Cu(100) at 100 K. B. EELS of S02/Cu(100). Figure 6 shows the EEL spectra of a saturation coverage of SO2on Cu( 100) exposed and measured

6168

The Journal of Physical Chemistry, Vol. 93, No. 16, 1989

r

+yph

SO,/CU( 100) specular

r

(b)

f JI

v

100)

SO,/CU(

m C 3

Leung et al.

300K

( c ) annealed 500K/300s

x .c_ m c

I

(b)

a,

c

annealed 350K/300s

C

a,

.c_

0

a,

K

500 1000 1500 Energy Loss (cm-')

0

2000

Figure 6. EEL spectra of a saturation coverage of SO2 on Cu(100) exposed and recorded at (a) 100 and (b) 300 K. The fwhm's of the

elastic peaks and the peak locations are indicated in cm-'. SO,/Cu(

100)

300K

1

Energy

.-cL1 C 3

e v

x

5 m c a,

c C

a,

.-c

-0 a,

iLi 300K (b)

\

2000

1000

Loss

Figure 7. EEL spectra of SOz on Cu(100) at 300 K at different coverages. The number of seconds of exposure using an effusive beam dosing method is indicated in each spectrum. The spectrum corresponding to a saturation coverage (-20 langmuirs) is given in (d).

in the specularly reflected direction at two different temperatures. Table I1 summarizes plausible assignments of the observed frequencies, compared with those of free SO2 molecules and other SO,-related systems. Except for the low-frequency region (GOO cm-I), two markedly different vibrational spectra was observed. By comparing with the gas-phase data,35the peaks at 1290, 978, and 5 15 cm-l in the 100 K spectrum (Figure 6a) can be assigned to asymmetric stretch, symmetric stretch, and bending modes, respectively. The two S - 0 stretching frequencies are lower than the corresponding gas-phase data (cf. 1362 cm-l for asymmetric stretch and 1 151 cm-I for symmetric stretch) while the bending frequency at 515 cm-' is almost identical with the corresponding gas-phase data.35 The remaining broad band is treated as an envelope of two close-lying peaks at 346 and 290 cm-I. The feature at 346 cm-' corresponds closely to the S-Cu stretching frequency of S/Cu(lOO)* (cf. 336 cm-I) while the peak a t 290 cm-' may be interpreted as either a torsion mode or hindered translations of physiadsorbed SO2. The low-frequency part of the room temperature spectrum (Figure 6b) can be assigned in an analogous fashion as the 100 K spectrum. Three different peaks are observed at 859, 613, and 482 cm-I. W e postpone discussion of the assignments of these vibrations until later. It should also be noted that the width of the elastic peak changes significantly, from 44 cm-' fwhm at 300 K to 60 cm-' fwhm at 100 K. The peakbroadening effect is an indication of surface disorder arising from physiadsorption. This interpretation is supported by an increase

121 H2S

296 613 ,340 I

(a) saturation

0

(cm-')

add

482

LE

Energy

(cm-'1

300 K after different temperature treatments. The annealing temperature (in K) and time (in seconds) are indicated in (b) and (c). h

0

Loss

Figure 8. EEL spectra of a saturation coverage of SOz on Cu(100) at

500 1000 1500 Energy Loss (cm-')

2000

Figure 9. EEL spectra of a saturation coverage of SOz on Cu(100) at

300 K before (a) and after (b) exposing an additional 12 langmuirs of

HIS. in the diffuseness in the observed ~ ( 2 x 2 LEED ) pattern of the sample at 100 K compared to that a t 300 K. Figure 7 shows the vibrational spectra of SO2 on Cu(100) at 300 K as a function of coverage obtained using a beam dosing p r o ~ e d u r e . The ~ ~ *coverages ~~ of the lo-, 20-, and 30-s spectra 2/3, and 1 monolayer, respectively. correspond approximately to There is very little change in the spectrum except for the structure at 859 cm-I, which increases proportionally with coverage. A saturation coverage of SO, on Cu( 100) at 300 K was annealed to 350 and 500 K for 5 min using electron beam bombardment from the back of the sample. Drastic changes were observed in the vibrational spectra, shown in Figure 8. The low-temperature anneal to 350 K completely removed the highfrequency features at 613 and 859 cm-' and sharpened the lowfrequency band at 324 cm-'. Further annealing to 500 K did not produce any noticeable change. There is also some evidence of a weak shoulder near 172 cm-' after the anneal. By comparing the annealed spectra with that of S/Cu( 100): it is clear that the 172-cm-' feature corresponds to a phonon vibration while the feature at 324 cm-' is due to S-Cu stretch (Figure 8b,c). Finally, a saturation coverage of SO2on Cu(100) at 300 K was exposed to 12 langmuirs of H2S using an ambient dosing procedure. The EEL spectrum shown in Figure 9 indicates a very similar vibrational profile to the spectra of the annealed sample shown in Figure 8. In particular, the high-frequency features are not observed and the feature at 329 cm-' can again be attributed to S-Cu stretch.* Like the annealed spectra (Figure 8), the weak

EELS of H2S and SO2 on Cu(100) shoulder at 161 cm-l provides some evidence of a possible phonon vibration.

IV. Discussion A . Adsorption of H# on Cu(100). For C, triatomic molecules adsorbed on surfaces, up to nine vibrational modes may become infrared active. These include three normal modes and six additional modes including three hindered rotations and three hindered translations. The assignments of the observed surface vibrations of H2S on Cu( 100) are consistent with both the gasphase data35 and the earlier EELS data of S/Cu(100).8 In particular, the S-Cu stretch (at 320 cm-I) is identified by comparison with the reported S/Cu(lOO) EELS study8 (Figure la). The assignments of the features at 1144 and 2469 cm-' to molecular H-S bending and stretching modes, respectively, are unambiguous. The vibration at 178 cm-' can be attributed to either phonon vibrations or hindered translations of molecularly adsorbed H2S. The latter case is supported by a coordinated increase (Figure 5) and decrease (Figure 4) in the intensity of this structure concomitant with the other molecularly adsorbed H2S features at 1144 and 2469 cm-I. The features at 430 and 550 cm-' are more difficult to interpret. Very similar EELS spectra of H2Son Ni( and on Pt( 111)21 were also reported. Both studies assigned the lower frequency peak to H-S stretch20 or H-S bend2' but gave two different assignments to the higher frequency, H-related feature analogous to the one observed at 550 cm-' in the present work. In particular, Baca et a1.20 assigned the higher frequency vibration to H-Ni stretch while Koestner et aLzl identified the corresponding peak to a second H-S bend. These two studies also reported similar temperature-dependent adsorption behaviors of H2S as in the present work, with the higher (lower) frequency vibrations of the respective systems becoming stronger (weaker) at higher temperature. In the case of H2S/Pt( 11l), however, an earlier study by Baro et al.36identified a fairly strong low-frequency mode due to H-Pt asymmetric stretch with H adsorbed in a hollow site. This feature persisted in the spectra if the sample was annealed to 220 K, in direct analogue with the strong adsorption of the higher frequency feature observed by Koestner et aL2' The assignment of Baca et al. (Le., to H-metal stretch) therefore appears to be more consistent with the data of both Pt(ll1) and the present case. The assignments of the features at 550 and 430 cm-l to a H-Cu stretch with H in a stable (probably 4-fold hollow) site and to H-S vibrations (bend or stretch) of a sulfhydryl species, respectively, are in accord with the observed changes in the relative intensities of the angle-dependent spectra (Figure 3). The existence of atomic H on Cu( 100) is also consistent with an earlier study which suggested that the adsorption of H2 on Cu(100) involved atomic chemisorption and that activated dissociation was involved.) The reported EELS study of H 2 on C U ( ~ O Oindicated )~ that H 2 adsorbed molecularly only below 10 K. It should be noted that we cannot, however, rule out an alternative assignment of the feature at 430 cm-l to hindered rotations of molecularly adsorbed H2S. Nevertheless, judging from the small intensity of this peak and the absence of the feature at 178 cm-' at low coverage (Figure 5a,b), which we earlier suggested to be due to hindered translations, this assignment appears to be rather unlikely. Both FisherI8 and Baca et a1.20proposed a similar model for the adsorption of H2S on Ru(l10) and Ni(100), respectively, at liquid nitrogen temperature. In this model, complete irreversible dissociation of H2S into S(ads) and H(ads) at low coverage is followed by partial dissociation into H-S(ads) and H(ads) at intermediate coverage. The H(ads) may recombine to from H2 which is then desorbed from the surface. Finally, molecular adsorption of H2S is observed at higher coverage, most likely in the form of islands. This model appears to be consistent with the present EELS data of H2S on Cu(100) at 100 K, as evidenced by the coverage-dependent study (Figure 5). Without a detailed calculation of the vibrational spectrum assuming different surface structures, it is difficult to determine the bonding sites or geom(36) Baro, A. M.; Ibach, H.; Bruchmann, H. D. Surf. Sei. 1979,88, 384.

The Journal of Physical Chemistry, Vol. 93, No. 16, 1989 6169 etries of the observed surface species, S, H, SH, and H2S, from the EELS spectra. The initial dissociation of H2Sinto S and H at low coverage probably leaves the dissociated species in the most stable (4-fold hollow) site which leads to perpendicular surface vibrations. The similarity between the S-Cu stretching frequency in the present case and that of the S/Cu(lOO) case8 confirms this model. The small value of the H-Cu stretching frequency also suggests that H is adsorbed in a 4-fold site although as pointed out by Baca et al.20one cannot unambiguously identify the site based only upon the magnitude of the vibrational frequency. At intermediate and higher coverage, it is not clear from the frequencies of the weakly bonded S H and H2Sspecies what bonding sites that these species may occupy. In fact the nature of the H-S vibration (Le., whether it is a stretching or bending mode) is not clear. However, the intensity variation of this feature as a function of scattering angle (Figure 3) indicates a strong dipole contribution. For the SH species it is therefore more likely, though not conclusively, that the H-S stretching mode is being observed. For the molecular H2S species, there are similar difficulties in identifying the adsorption sites, although the associated frequencies allow us to distinguish between the stretching and bending modes. Unlike H20,5,6for which molecular adsorption on metal surfaces appears to be the rule, the dissociation of H2Sand the formation of sulfhydryl (SH) species appear to be quite common on most metal surfaces studied to date. Except for W(100)19 where it occurred at rmm temperature, SH formation occurred below 170 K on most metal surfaces, including Rh(100),13 Ru(1 10),I8 Ni( Pt( l l 1),21and Cu( 100). Like molecularly adsorbed H2S, the S H species binds weakly to the surface and desorbs easily. This behavior can be attributed to the much smaller H-S bond energy in H2S when compared with the H - O bond energy in H20. The desorption behavior of H2Son Cu( 100) is also very similar to that observed on the other metal surface^,^^-'^^^^*^' where H2 and H2S appear to be the only desorption products. In particular, thermal desorption studies of HIS on Pt(l1 1)21and on Rh( 10O)l3 suggested two molecular H2S desorption states: one due to chemisorbed H2Sinteracting with the sulfided metal surfaces and the other due to H2Sweakly bonded with islands of chemisorbed H2S. This picture of weak H2S adsorption is consistent with the temperature-dependent EELS study in the present work. B. Adsorption of SO2 on Cu(J00). In the case of SO2 adsorption, the three high-frequency features at 100 K (Figure 6a) can be assigned to the corresponding normal modes of free SO2 molecules by a comparison with the gas-phase data.35 (See Table 11.) The low-frequency features at 346 and 290 cm-' are most likely due to S-Cu stretch and hindered translations, respectively. The similarity of the observed vibrational features with those of suggests that SO2 was free SO2 molecules35 and S/Cu( physisorbed on the surface at 100 K. A similar adsorption pattern was also reported for SO2on CsCl films.27 In particular, associative adsorption on CsCl was observed for three different adsorption phases involving different temperature ranges: a phase (220-240 K), /3 phase (146-205 K), and y phase (>200 K). The characteristic frequencies corresponding to these phases are given in Table 11. In the case of SO2on C S C ~only , ~ ~one low-frequency peak, presumably due to S-CsCl stretch, was reported although there were significant backgrounds in these infrared spectra. The assignment of the feature at 290 cm-' in the SO2on Cu( 100) case to hindered translation is therefore not unique. Other possibilities include (1) a torsion mode and (2) two different adsorption sites for SO2molecules which give rise to two possible S-Cu vibrational frequencies. The EEL spectrum of 12 langmuirs of SO2on Pd(100) recorded at 115 K by Burke and Madix30was very similar. In both cases, the frequencies of the observed (molecular) features compare favorably with the corresponding ones for solid S02.38 The spectral features of the room temperature spectrum (Figure 6b) are more difficult to interpret. The low-frequency features (290 and 340 cm-') at 300 K could be assigned in a similar fashion (37) Natamato, K.; McCarthy, P. J. Spectroscopy and Structure of Metal Chelate Compounds; Wiley: New York, 1968, p 267. ( 3 8 ) Anderson, A,; Walmsley, S.H. Mol. Phys. 1966, I O , 391.

6170

The Journal of Physical Chemistry, Vol. 93, No. 16, 1989

as for the low-temperature spectrum. No high-frequency peaks (> 1000 cm-I), characteristic of the SO, internal vibrations that were observed in the low-temperature case, were found. The three new features observed at 482, 613, and 859 cm-l were indicative of different surface species (other than SO,) resulting from the room temperature adsorption. Previous studies of SO, on Fe and Ni,22Ag Pt(l1 Pd( and p(2X2)S/Pd( all suggested dissociative adsorption and reported evidence of SO4 and S as the principal adsorption products, particularly at higher temperatures. Other studies of SO2on Ca025and Mg0,26however, suggested SO, as a possible surface species upon adsorption. The EELS study of the oxidation of SO, on Ag( 110) reported by Outka et aL3' suggested the presence of SO, and SO4 at different temperatures. From Table 11, the strong peaks at 859 and 613 cm-I and the weak feature at 482 cm'.' can be tentatively assigned respectively to symmetric/asymmetric S-0 stretch, symmetric S-0 bend, and asymmetric S-0 bend of SO,. The band at 859 cm-I is broad and can be attributed to two overlapping bands. Earlier infrared data relating the SO, (x = 3 or 4) vibrations suggested that in general the asymmetric S-O stretching frequency is greater than 1100 cm-' for SO4but smaller than 1100 cm-l for SO,.'' The relatively low value of the S-0 stretching band at 859 cm-I therefore strongly favors the presence of SO, rather than SO4. The sulfite group, SO,, can coordinate to the metal surface as an unidentate or bidentate ligand. Both of these bonding arrangements were considered for the case of SO3 on Ag( 110) by Outka et al.,31who concluded from their EEL spectra that unidentate 0 bonding occurred between 200 and 325 K while bidentate bonding was found between 325 and 500 K. From the infrared data of the coordination complexes compiled by Outka et al.,,' 0 coordination with SO, in a pyramidal arrangement (see Table 11) appears to be most likely for Cu(100). This is consistent with Nyberg and Larsson's observation that a strong vibration above 975 cm-' is indicative of S c o ~ r d i n a t i o n . In ~ ~the case of bidentate geometry, there are at least four possible cases, each involving bonding of a S-0 pair or an 0-0 pair to either one or two metal atoms. From the infrared studies of bidendate coordination complexes of the type [ C O ( ~ ~ ) ~ S O ,where ] X , X = C104-, NO,, I-, or SCN-, it was found that the spectrum would be more complex due to the existence of many different vibration^.^' The simplicity of our spectrum further supports that unidentate coordination through the 0 atom represents the most probable bonding geometry. From the observed vibrational spectrum at 300 K, we have already concluded that the reaction SO, + SO2 S (abs) + SO, (abs) involving the production of SO4is unlikely. Instead, surface reactions generating SO, and S (and/or 0)as adsorbed surface species are considered. We therefore propose the following reaction scheme to account for the observed adsorption pattern. SO2 SO(abs) O(abs) (1) SO(abs) + SO, S(abs) + SO,(abs) (2) O(abs) + SO, SO,(abs) (3) SO(abs) S(abs) O(abs) at higher coverage or above 350 K (4)

-

-+

-- + -

(39) Nakamato, K. Infrared Spectra of Inorganic and Coordination Compounds; Wiley: New York, 1963. (40) Nyberg, B.; Larsson, R. Acta Chem. Scand. 1973, 27, 63.

Leung et al. This model (reactions 1-3) is certainly not unique in accounting for the observed S(abs) and S03(abs) species. It emphasizes dissociative adsorption of SO2 into SO(abs) and O(abs) (reaction I ) , which subsequently combines with SO2 to give S03(abs) (reaction 3). The reaction of SO(abs) with SO, gives the products S(abs) and SO,(abs) (reaction 2). As evident from the coverage-dependent spectra (Figure 7), the increase in the intensity of the S-0 stretching modes but not the bending modes at higher coverages suggests the build-up of additional species other than SO,. In this model, the most likely candidate is SO itself. Furthermore, additional dissociation of SO into S and 0 (reaction (4)) cannot be ruled out. It should be noted that both S-Cu8 and 0-Cu9 stretching modes occur at the same frequency, and it is therefore impossible to differentiate between the two. The two low frequency features at 296 and 340 cm-' can also be explained by the presence of two different S-Cu (stretching) vibrations corresponding to Cu-SO or Cu-SO,. This implies that both SO and SO, are present at all coverages and SO continues to build up after the saturation of SO, on the surface. With the present resolution of 40 cm-' fwhm, it is not possible to resolve these two features and investigate their intensity variations as a function of coverage. The surface species (SO or SO,) appear to be weakly adsorbed because only a small temperature rise is required to remove most of the vibrational features associated with these species (Figure 8). After annealing, the low-frequency band sharpens and can again be attributed to a S-Cu (or O-cu) stretch. As well, the possibility of dissociation of SO and SO, to S and/or 0 cannot be ruled out. C. Reaction of H2S with Preadsorbed SO, on Cu(100). Except for the increase in the intensity of the 329-cm-' peak, an exposure of 12 langmuirs of H2S to preadsorbed SO, on Cu(100) at room temperature produced a spectrum almost identical (Figure 9) with that of the annealed sample (Figure 8). We suggest above that SO, is the most likely surface species upon adsorption of SO, on Cu( 100) at room temperature. The increase in S-Cu stretch in the resulting vibration spectrum (Figure 9b) suggests the formation of S(abs) although it is not possible to identify the reactions involved definitively without other experimental data such as thermal desorption. This spectrum also shows that any reaction involving the formation of SH(abs) is unlikely because the characteristic vibration at 430 cm-I is missing. The SH species can, however, act as an intermediate. One possible reaction scheme would involve the dissociation of H2S to H and SH, which would then undergo further dissociation to H and S(abs). The dissociation of the SH species at room temperature is consistent with the earlier temperature-dependent study shown in Figure 4. The resulting H may combine with SO, to form H,SO,. Further study is required to substantiate this proposed mechanism.

Acknowledgment. This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Science Division of the U S . Department of Energy under Contract No. DE-AC03-76SF00098. K.T.L. gratefully acknowledges support from the Natural Sciences and Engineering Research Council of Canada and the Killam Memorial Foundation. Registry No. Cu, 7440-50-8; H2S, 7783-06-4; SO*, 7446-09-5.